ANNUAL SUMMARY Eastern North Pacific Hurricane Season of 1999

1036 MONTHLY WEATHER REVIEW VOLUME 132

ANNUAL SUMMARY

Eastern North Paci®c Hurricane Season of 1999

JOHN L. BEVEN II AND JAMES L. FRANKLIN Tropical Prediction Center, NWS, NOAA, Miami, Florida

(Manuscript received 11 February 2003, in ®nal form 20 October 2003)

ABSTRACT The 1999 hurricane season in the eastern North Paci®c is summarized, and individual tropical storms and hurricanes are described. Producing only nine named storms, the season tied 1996 as the second least active on record. Hurricane Dora was the strongest and longest-lived cyclone of the season. Hurricane Greg, the only cyclone to make landfall during the season, weakened to a tropical storm just before moving ashore in Baja California, Mexico. Fifteen deaths resulted from the tropical cyclones.

1. Introduction central Paci®c basin, with Dora later crossing the in- The eastern North Paci®c hurricane basin covers the ternational date line into the western North Paci®c ba- area north of the equator and east of 140ЊW longitude sin. to Central and North America. Nine tropical storms One can speculate on the reasons for the inactive formed in this basin during the 1999 hurricane season. season. One possibility is the active 1999 Atlantic sea- Of these, six became hurricanes and two became major son (Lawrence et al. 2001), which featured 16 tropical hurricanesÐcategory 3 or higher (maximum sustained cyclones (TCs). Climatologically, most eastern North winds of 97 kt or greater) on the Saf®r±Simpson Hur- Paci®c cyclones can be traced to disturbances (tropical ricane Scale (Simpson 1974); the 1966±98 averages waves) that cross Central America from the Caribbean are 16 tropical storms, 9 hurricanes, and 4 major hur- (Simpson et al. 1969; Avila and Guiney 2000). An ricanes. There were ®ve additional tropical depressions increase in the number of Atlantic storms could po- that failed to reach tropical storm intensity. The 1999 tentially decrease the vigor of waves crossing into the season ties the 1996 season (May®eld and Rappaport Paci®c. Atlantic Hurricanes Cindy, Dennis, Floyd, 1998) for the second least active since satellite obser- Gert, and Jose all formed from tropical waves and then vations began in 1966. Only the 1977 season, with moved out of the Tropics. While tropical waves spawn eight tropical storms (Gunther 1978), was less active. more than one TC occasionally, the remnant wave vor- Table 1 lists the tropical storms and hurricanes of ticity after genesis often appears greatly reduced. A 1999, while Fig. 1 shows a map of their tracks. Most pattern of above-normal activity in the Atlantic cou- of the data used in determining the tracks and inten- pled with below-normal activity in the eastern Paci®c sities (1-min average sustained winds at 10 m) are de- was also noted in 1995, 1996, and 1998. Conversely, rived from satellites; these include Dvorak (1984) in- the active 1997 season in the eastern Paci®c was ac- tensity estimates, cloud-drift winds, and ocean surface companied by a below-normal number of Atlantic sys- scatterometer data (e.g., Tsai et al. 2000). Other sources tems. This inverse relationship is intriguing, but does of data include infrequent surface, upper-air, and radar not occur every year, as noted by Lander and Guard observations. There were no reconnaissance ¯ights into (1998). eastern North Paci®c tropical cyclones in 1999. All 1999 eastern Paci®c tropical cyclones (including Hurricane Greg, which crossed the southern tip of the nondeveloping tropical depressions) could be as- Baja California, Mexico, as a tropical storm on 7 Sep- sociated with Atlantic tropical waves that crossed Cen- tember, was the only storm to make landfall in 1999. tral America. However, the background monsoonlike Hurricanes Dora and Eugene crossed 140ЊW into the environment may have played a role in the develop- ment of several cyclones. In some cases, the wave de- Corresponding author address: Dr. John L. Beven II, National veloped into a tropical cyclone upon entering the fa- Hurricane Center, 11691 SW 17th Street, Miami, FL 33165-2149. vorable background environment. In other cases, the E-mail: [email protected] wave may have assisted the development of a preex-

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC APRIL 2004 ANNUAL SUMMARY 1037

TABLE 1. Eastern Paci®c tropical storms and hurricanes of 1999. Max 1-min Min sea level Name Class* Dates** wind speed (kt) pressure (mb) Deaths Adrian H 18±22 Jun 85 973 6 Beatriz H 9±17 Jul 105 955 Calvin T 25±27 Jul 35 1005 Dora H 6±23 Aug 120 943 Eugene H 6±15 Aug 95 965 Fernanda T 17±22 Aug 55 994 Greg H 5±9 Sep 65 986 9 Hilary H 17±21 Sep 65 987 Irwin T 8±11 Oct 50 997

* T: tropical storm, max sustained winds 34±63 kt. H: hurricane, max sustained winds 64 kt or higher. ** Dates based on UTC and include tropical depression stage. isting monsoon disturbance. Unfortunately, the avail- of 4 and 2, respectively. While the mean ITCZ appears able data often do not allow for a distinction between weak in July (Fig. 3b), monthly low-level wind anom- the two possibilities. alies (not shown) suggest it was close to normal Figure 2 shows an example of a tropical wave cross- strength. Also, 200±850-mb wind shear data (not ing Central America that eventually became Hurricane shown) indicate no obvious inhibiting factors. Thus, Greg. In this particular case, a monsoonlike environ- the reason for the inactive July remains a mystery. ment with some disturbed weather was in place before During August, a well-de®ned mean ITCZ circula- the arrival of the wave. However, tropical cyclogenesis tion was centered southwest of Baja California (Fig. did not occur until the wave arrived. Figure 2 suggests 3c); it also strongly appeared in monthly wind anom- that the tropical cyclogenesis showed better spatial and alies (not shown). Hurricane Eugene and Tropical temporal continuity with the preexisting disturbed Storm Fernanda formed in the vicinity of this feature, weather, which in turn suggests that the wave acted as while Hurricane Dora became a major hurricane in that a trigger for the development of the preexisting mon- area. soon disturbance. During September, the ITCZ shifted northward to Another possible explanation for the reduced activ- near the coast of Mexico east of 105ЊW (Fig. 3d), cre- ity is that the sea surface temperature (SST) pattern ating a strong cyclonic anomaly along 18Њ±20ЊN from combined with the location of the intertropical con- 115ЊW eastward into Mexico (not shown). This shift vergence zone (ITCZ) to create a smaller than normal resulted in disturbances passing over or near the moun- area favorable for tropical cyclone formation. Colder tainous areas of southern Mexico, likely limiting cy- than normal SSTs along the equator associated with La clogenesis. Hurricanes Greg and Hilary formed on the NinÄa, combined with the normal cold SSTs generally southeast side of the mean cyclone centered southwest north of 20ЊN, may have limited the area of warm water of Baja California. necessary for tropical cyclone development and main- tenance. This was most notable during August west of 120ЊW, where only a 10Њ±12Њ-wide east±west band of 2. Tropical storm and hurricane summaries SSTs warmer than 26ЊC existed (not shown). a. Hurricane Adrian: 18±22 June Figure 3 shows monthly mean 850-mb winds for June, July, August, and September 1999. Note the A persistent area of disturbed weather located south month-to-month variation in the location and intensity of the Gulf of Tehuantepec developed a low-level cir- of the ITCZ. During June, the mean position extended culation and convective banding on 16 June. This oc- from Panama west-northwestward to a few hundred curred when a tropical wave, which had moved off the miles southwest of Baja California with mean circu- west coast of Africa 11 days earlier, moved into the lation centers near the western end and just south of preexisting disturbed area. Further development fol- southeastern Mexico. Both circulation centers are lowed, producing a tropical depression about 225 n mi anomalous (not shown). The western center, while ap- southeast of Acapulco, Mexico, near 0600 UTC 18 pearing favorable for TC development, is located near June (Fig. 1). The depression became Tropical Storm cold SSTs. This superposition may have helped sup- Adrian later that day. A deep-layer ridge over Mexico press activity, as any nascent TC in this area would helped steer the cyclone west-northwestward on a track likely have encountered cold water before genesis parallel to the coast of Mexico through 20 June. Adrian could occur. became a hurricane on 20 June and maintained an es- Five tropical cyclones formed in July, which is near timated peak intensity of 85 kt from 1800 UTC 20 normal. However, only 2 of these became storms and June to 0000 UTC 21 June. An eye was brie¯y seen only 1 a hurricane, numbers that are half of the normal in satellite imagery during this time (not shown).

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC 1038 MONTHLY WEATHER REVIEW VOLUME 132 . 1. Tracks of eastern Paci®c tropical storms and hurricanes during 1999. IG F

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC APRIL 2004 ANNUAL SUMMARY 1039

FIG. 2. Sequence of daily Geostationary Operational Environmental Satellite-8 (GOES 8) infrared satellite images at 1200 UTC from 27 Aug to 4 Sep 1999. Dates are indicated on the right-hand side of the ®gure. Dashed line marks the tropical wave that eventually triggered Hurricane Greg (G).

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC 1040 MONTHLY WEATHER REVIEW VOLUME 132

FIG. 3. The 850-mb mean monthly winds for (a) Jun, (b) Jul, (c) Aug, and (d) Sep 1999. Arrow length is proportional to the wind speed (m sϪ1). Labeled dots show the genesis locations of tropical cyclones. Thick black line is the mean position of the ITCZ inferred from winds. Wind data courtesy of the National Oceanic and Atmospheric Administration Climate Prediction Center.

As the cyclone passed near Socorro Island, Mexico, pression on 22 June. Adrian dissipated later that day, later on 21 June, southeasterly vertical wind shear and about 300 n mi southwest of Cabo San Lucas, Mexico. cooler SSTs caused Adrian to weaken to a tropical storm. Limited observations in Adrian include a report of 34- Adrian turned toward the west and weakened to a de- kt winds and a 998.6-mb pressure on 19 June from the

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC APRIL 2004 ANNUAL SUMMARY 1041

FIG.3.(Continued) ship L'atalante. Socorro Island reported 40-kt winds and b. Hurricane Beatriz: 9±17 July a 993-mb pressure at 1200 UTC 21 June. Although Adrian did not make landfall, outer rain- Beatriz formed from a tropical wave that entered the bands spread over western Mexico causing two deaths eastern tropical Atlantic on 26 June. The wave moved from ¯ash ¯oods. Four other deaths were attributed to over Central America and extreme eastern Mexico on high surf. 4±5 July. Disturbed weather consolidated in the vicinity

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC 1042 MONTHLY WEATHER REVIEW VOLUME 132 arch Laboratory. . 4. SSM/I 85-GHz brightness temperature images of Hurricane Beatriz at (left) 0147 and (right) 1701 UTC 12 Jul 1999. Imagery provided by the Naval Rese IG F

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC APRIL 2004 ANNUAL SUMMARY 1043 of the wave in the Gulf of Tehuantepec on 6 July, and though northeasterly vertical shear provided a less than a cyclonic circulation was evident the next day. Further ideal environment, the cyclone became Tropical Storm development was slow, and it was 0600 UTC 9 July Dora on 7 August. Decreasing shear on the next day before a tropical depression formed about 300 n mi allowed Dora to become a hurricane. Intensi®cation con- south of Lazaro Cardenas, Mexico (Fig. 1). The de- tinued, with Dora reaching an estimated peak intensity pression moved west-northwestward and became a trop- of 120 kt near 0000 UTC 12 August (Fig. 5). After ical storm later that day. Beatriz became a hurricane on weakening brie¯y, perhaps in association with a con- 11 July while a midtropospheric ridge to its north helped centric eyewall cycle inferred from SSM/I and infrared steer the cyclone westward. Intensi®cation stopped tem- satellite imagery (not shown), Dora regained 120-kt poraily on 12 July; Special Sensor Microwave Imager winds near 1800 UTC that day. Although slow weak- (SSM/I) data (Fig. 4) suggest this was most likely due ening followed the second peak, Dora still had estimated to an eyewall replacement cycle. Once this cycle was 100-kt winds when it crossed 140ЊW longitude into the complete, a well-de®ned 20 n mi diameter eye formed, central Paci®c basin. and the hurricane reached an estimated peak intensity Analyses from the Central Paci®c Hurricane Center of 105 kt on 13 July. and the Joint Typhoon Warning Center show that Dora A weakness in the subtropical ridge north of Beatriz then continued westward across the central Paci®c, pass- allowed a west-northwestward motion over gradually ing south of Hawaii and crossing the international date cooler waters on 13 July. This resulted in a gradual line into the western North Paci®c as a tropical storm weakening that lasted through 15 July. More rapid on 20 August. Dora weakened to a depression on 22 weakening on 16 July reduced Beatriz to a tropical storm August and dissipated the next day about 550 n mi and then to a depression. The cyclone turned westward north-northwest of Wake Island, a full month after its and dissipated the next day, about 1100 n mi east of seedling disturbance crossed the African coast. the Hawaiian Islands. Dora was the longest-lived and strongest eastern Pa- The only signi®cant surface observation was from the ci®c tropical cyclone of 1999 and was a major hurricane ship Belo Oriente, which reported 35-kt winds about for 4 days. It was also the ®rst eastern Paci®c tropical 190 n mi north-northeast of the center at 1200 UTC 13 cyclone to spend time in all three Paci®c tropical cy- July. clone basins since Hurricane John did so in 1994 (Pasch and May®eld 1996). c. Tropical Storm Calvin: 25±27 July A tropical wave that emerged from the west coast of e. Hurricane Eugene: 6±15 August Africa on 9 July crossed Central America into the Paci®c Ocean on 20 July. The wave moved into a convectively Eugene had its origin in a tropical wave that crossed active ITCZ, with a midlevel circulation developing on Central America on 28±29 July and moved westward 23 July. Additional development produced a tropical for several days with little sign of organization. Con- depression about 560 n mi southwest of Cabo San Lucas vection associated with the wave became more consol- by 0600 UTC 25 July (Fig. 1). Although the low-level idated on 5 August, and the system became a tropical center was displaced northwest of the primary convec- depression near 0600 UTC the next day about 850 n mi tion, the depression became a tropical storm 6 h later southwest of Cabo San Lucas (Fig. 1). The cyclone while it moved west-northwestward. West-northwest- moved west-northwestward and became a tropical storm erly vertical wind shear prevented further intensi®cation later that day. A strengthening ridge to its north helped and caused Calvin to weaken to a depression on 26 July. turn Eugene generally westward on 8±9 August, allow- Calvin turned to the northwest, and this track continued ing the cyclone to stay over warm water and intensify. until the cyclone dissipated about 750 n mi west-south- Eugene became a hurricane on 8 August and developed west of Cabo San Lucas on 27 July. a 10 n mi diameter eye. It reached an estimated peak intensity of 95 kt the next day. This occurred near 133ЊW, which is signi®cantly farther west than the cli- d. Hurricane Dora: 6±23 August matological location of peak intensity for eastern Paci®c A tropical wave that crossed the West Africa coast tropical cyclones. Slow weakening ensued, and Eugene on 23 July emerged into the eastern Paci®c basin on 3± crossed into the central Paci®c near 0900 UTC 11 Au- 4 August. The system became better organized the next gust with 75-kt winds. day and developed into a depression about 290 n mi Eugene continued generally westward across the cen- south of Acapulco near 0000 UTC 6 August (Fig. 1). tral Paci®c and slowly weakened, becoming a tropical The cyclone moved west-northwestward through 8 Au- storm on 13 August. A west-southwestward motion on gust; subsequently, a mid- to upper-level ridge building 14 August brought Eugene to a region of increased ver- to the north of the cyclone contributed a westward mo- tical wind shear, and the cyclone weakened to a tropical tion (a motion that would continue until the cyclone depression. Eugene dissipated the next day about entered the central Paci®c basin on 14 August). Even 500 n mi southeast of Johnston Island.

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC 1044 MONTHLY WEATHER REVIEW VOLUME 132

FIG.5.GOES-10 visible image of Hurricane Dora at 0100 UTC 11 Aug 1999. Imagery provided by the Cooperative Institute for Meteorological Satellite Studies, University of WisconsinÐMadison.

f. Tropical Storm Fernanda: 17±22 August g. Hurricane Greg: 5±9 September

The tropical wave that spawned Fernanda emerged Hurricane Greg developed, in part, from the same from the African coast on 1 August. Convection asso- tropical wave that triggered Tropical Storm Emily in the ciated with the wave increased when it moved into the Atlantic east of the Lesser Antilles on 24 August (Lawr- ence et al. 2001). After spawning Emily, the southern eastern North Paci®c basin on 10 August. Slow and portion of the wave continued westward and crossed gradual development occurred while the wave continued Central America during 31 August and 1 September westward over the next several days. A low-level cir- (Fig. 2). Thunderstorm activity became concentrated culation within the system became better de®ned on 16 while cloud-banding featured developed on 3±4 Sep- August, and scatterometer winds from the European tember, and the system became a tropical depression Remote Sensing Satellite-2 (ERS-2) indicated that a trop- near Manzanillo, Mexico, at 1200 UTC 5 September ical depression formed near 0600 UTC 17 August, about (Fig. 1). It is of interest that the combination of the 400 n mi south-southwest of Socorro Island (Fig. 1). favorable monsoonal environment and tropical wave The depression moved west-northwestward becoming a was also responsible for the genesis of Tropical De- tropical storm the next day. Fernanda reached an esti- pression 7 in the western Gulf of Mexico on 5 Septem- mated peak intensity of 55 kt on 19 August. Fernanda ber. Thus this wave was involved in the genesis of three turned west-southwestward, a track that kept the cyclone tropical cyclones. over warm SSTs but took the system into an environ- A report of 42-kt winds and a pressure of 1006.5 mb ment of increasing easterly vertical wind shear. Fernan- at 1800 UTC 5 September from the ship Hume Highway da weakened to a depression on 21 August and dissi- indicated the depression had become Tropical Storm pated the next day about 1300 n mi west-southwest of Greg by that time. While Greg moved on a general Cabo San Lucas. northwestward track very close to the coast of Mexico,

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC APRIL 2004 ANNUAL SUMMARY 1045

TABLE 2. A homogeneous comparison of of®cial and CLIPER track forecast errors (n mi). Forecast period (h) 0 1224364872 1999 Avg of®cial error 8.6 32.2 60.1 83.2 103.7 135.2 1999 Avg CLIPER error 8.6 35.1 70.2 106.5 143.0 215.9 No. of cases 128 128 114 102 90 69 1989±98 Avg of®cial error 12.0 37.8 70.5 103.8 135.4 194.5 1989±98 Avg CLIPER error 12.0 40.1 76.8 117.2 157.0 228.7 No. of cases 2580 2574 2308 2035 1786 1367 a large area of very deep convection formed near the Baja California. Northeasterly vertical wind shear pre- center. Radar from Los Cabos, Mexico, suggested the vented further intensi®cation, and no signi®cant change formation of a ragged eye during the morning of 6 Sep- in strength occurred for the next 24 h. Thereafter, Greg tember. Figure 6 shows a visible satellite image of Greg weakened to a tropical storm as it made landfall near near this time. Greg became a hurricane at 1800 UTC Cabo San Lucas around 2100 UTC 7 September. Greg that day while it approached the southern portion of turned west-northwestward, then westward, and moved

FIG.6.GOES-10 visible image of Hurricane Greg at 1600 UTC 6 Sep 1999. Imagery provided by the Naval Research Laboratory.

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC 1046 MONTHLY WEATHER REVIEW VOLUME 132

TABLE 3. A homogeneous comparison of of®cial and SHIFOR max 1-min wind speed forecast errors (kt). Forecast period (h) 0 1224364872 1999 Avg of®cial error (bias) Ϫ0.5 0.0 Ϫ0.3 Ϫ2.5 Ϫ6.1 Ϫ11.4 1999 Avg absolute of®cial error 1.8 5.6 9.7 12.3 15.9 20.5 1999 Avg SHIFOR error (bias) Ϫ0.5 Ϫ2.2 Ϫ4.7 Ϫ9.3 Ϫ13.9 Ϫ25.8 1999 Avg absolute SHIFOR error 1.8 6.7 11.8 15.8 20.2 29.3 No. of cases 128 128 114 102 90 69 1990±98 Avg of®cial error (bias) Ϫ0.9 Ϫ1.3 Ϫ2.3 Ϫ3.9 Ϫ5.6 Ϫ5.9 1990±98 Avg absolute of®cial error 3.8 6.7 10.3 13.1 14.9 16.6 1990±98 Avg SHIFOR error (bias) Ϫ0.9 Ϫ3.0 Ϫ5.7 Ϫ8.9 Ϫ11.4 Ϫ15.5 1990±98 Avg absolute SHIFOR error 3.8 7.4 11.4 14.5 16.5 18.4 No. of cases 2369 2364 2126 1892 1665 1283 over cooler waters and further weakened. It became a ical Depression 11 in the Bay of Campeche on 4 October depression on 8 September and dissipated the next day (Lawrence et al. 2001), produced disturbed weather near about 150 n mi west-northwest of Cabo San Lucas. Acapulco on 5 October. Strong easterly vertical wind During Greg's passage close to the Mexican coast, shear slowed development until 8 October, when the San Jose del Cabo, on the southern tip of Baja Cali- system became a tropical depression about 125 n mi fornia, reported sustained winds of 35 kt with gusts to south-southwest of Manzanillo (Fig. 1). The cyclone 40 kt and a minimum pressure of 995 mb at 2100 UTC moved slowly north-northwestward and reached tropical 7 September. The cyclone produced torrential rains over storm strength later that day. Irwin was within 75 n mi portions of southwestern Mexico. Nearly 230 mm were of the coast of Mexico on 9 October when a strength- measured in Manzanillo, about 205 mm in Colima, and ening midtropospheric ridge forced it to turn west-north- 125 mm in the Islas Marias. westward. The storm reached an estimated peak inten- Media reports indicate that rains from Greg caused sity of 50 kt after this turn. A combination of north- extensive ¯ooding over the states of Colima, Jalisco, easterly vertical wind shear and entrainment of stable Michoacan, Nayarit, and Sinaloa, killing nine people. air caused Irwin to weaken on 10 October. It weakened No monetary damage ®gures are available. to a depression while passing near Socorro Island, and it dissipated the next day about 350 n mi southwest of h. Hurricane Hilary: 17±21 September Cabo San Lucas. The only known observation of tropical storm force The development of Hilary was associated with a winds was from the ship Lincoln Spirit, which reported tropical wave that crossed the West African coast on 29 45-kt winds at 1800 UTC 9 October. There were no August and reached the eastern Paci®c on 11±12 Sep- reports of damage or casualties. tember. The system moved slowly near the Mexican coast for several days while producing persistent dis- turbed weather. A low-level circulation formed by 17 3. Tropical depressions September, and the system became a tropical depression Five other tropical depressions occurred during the about 475 n mi south-southeast of Cabo San Lucas near 1999 season, with three in July and two in August. 0600 UTC that day (Fig. 1). While moving west-north- Tropical Depression 3-E occurred from 14±15 July. westward, it became a tropical storm the next day. Hi- Tropical Depression 4-E existed from 23±25 July and lary turned north-northwestward on 19 September and was close to tropical storm strength on 24 July. Tropical became a hurricane, with a banding-type eye and 65-kt Depression 6-E developed on 26 July about 360 n mi winds, the next day. Movement over cooler waters southwest of Tropical Storm Calvin and dissipated 2 caused Hilary to weaken to a tropical storm later that days later. Tropical Depression 9-E was observed from day and to a tropical depression on 21 September. The 13±15 August. Finally, Tropical Depression 11-E cyclone dissipated about 130 n mi west of central Baja formed on 23 August from a broad low pressure system California shortly thereafter. that likely resulted from a combination of a tropical The only observation of tropical storm conditions was wave and a background monsoon environment. Winds from the ship Salus, which reported 37-kt winds about were near tropical storm force at genesis. However, no 120 n mi from the center at 0000 UTC 20 September. further development occurred, and the system dissipated the next day. i. Tropical Storm Irwin: 8±11 October Irwin formed within a tropical wave that moved 4. Forecast veri®cation across Central America and eastern Mexico on 1±3 Oc- The Tropical Prediction Center (TPC) issues advi- tober. This system, which spawned Atlantic basin Trop- sories every6honalltropical cyclones in the Atlantic

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC APRIL 2004 ANNUAL SUMMARY 1047 and eastern North Paci®c hurricane basins. These ad- greater. The SHIFOR forecast errors were near the long- visories include 12-, 24-, 36-, 48-, and 72-h forecasts term averages at 12 and 24 h. They were 5%±10% great- of track and intensity. A ``best track'' position and in- er than the average at 36 h, 20% greater at 48 h, and tensity (maximum 1-min wind speed) are determined at 60%±70% greater at 72 h. The of®cial intensity forecast 6-h intervals for each tropical cyclone after the fact thus were skillful even at 72 h. using all available data. These best tracks (shown in Fig. 1) are the basis for verifying forecasts. Acknowledgments. Lixion Avila, Miles Lawrence, The track forecast error is de®ned as the great circle and Richard Pasch contributed to the writing of this distance between a forecast position and a best-track report, while Stephen Baig produced the track chart. position for the same time. Two intensity forecast errors Rick Knabb produced the monthly wind shear anoma- are calculatedÐthe absolute error, de®ned as the ab- lies. Satellite imagery shown in this report was provided solute value of the difference between the forecast and by the Cooperative Institute for Meteorological Satellite best-track wind speed, and the forecast bias, which is Studies at the University of WisconsinÐMadison and the Naval Research Laboratory. the algebraic difference between a forecast and a best- track wind speed. The skill of TPC forecasts is measured against the REFERENCES baseline of climatology and persistence forecasts. For Avila, L. A., and J. L. Guiney, 2000: Eastern North Paci®c hurricane track forecasts, the baseline is the forecast from the season of 1998. Mon. Wea. Rev., 128, 2990±3000. climatology±persistence (CLIPER) statistical track Dvorak, V. F., 1984: Tropical cyclone intensity analysis using satellite data. NOAA Tech. Rep. NESDIS 11, 47 pp. model (Neumann 1972). For intensity forecasts, the Gunther, E. B., 1978: Eastern North Paci®c tropical cyclones of 1977. baseline is the forecast from the Statistical Hurricane Mon. Wea. Rev., 106, 546±558. Intensity Forecast (SHIFOR) system (Jarvinen and Neu- Jarvinen, B. R., and C. J. Neumann, 1979: Statistical forecasts of mann 1979). tropical cyclone intensity. NOAA Tech. Memo. NWS NHC-10, 22 pp. Table 2 shows the average of®cial TPC track forecast Lander, M. A., and C. P. Guard, 1998: A look at global tropical errors for 1999 in the eastern North Paci®c, along with cyclone activity during 1995: Contrasting high Atlantic activity the 1999 CLIPER average errors, the NHC average fore- with low activity in other basins. Mon. Wea. Rev., 126, 1163± cast error for the previous 10 yr, and the average CLI- 1173. Lawrence, M. B., L. A. Avila, J. L. Beven, J. L. Franklin, J. L. Guiney, PER errors for the previous 10 yr. The 1999 of®cial and R. J. Pasch, 2001: Atlantic hurricane season of 1999. Mon. track forecast errors were 15%±25% smaller than the Wea. Rev., 129, 3057±3084. average of the previous 10 yr for all forecast periods. May®eld, M., and E. N. Rappaport, 1998: Eastern North Paci®c hur- ricane season of 1996. Mon. Wea. Rev., 126, 3068±3076. The 1999 CLIPER forecast errors were only about 10% Neumann, C. J., 1972: An alternate to the HURRAN tropical cyclone better than the previous 10-yr average, indicating that forecast system. NOAA Tech. Memo. NWS SR-62, 22 pp. much of the improvement in the of®cial forecasts was Pasch, R. J., and M. May®eld, 1996: Eastern North Paci®c hurricane due to increased forecaster skill. season of 1994. Mon. Wea. Rev., 124, 1579±1590. Simpson, R. H., 1974: The hurricane disaster potential scale. Weath- Table 3 shows the average bias and average absolute erwise, 27, 169±186. errors for TPC intensity forecasts. The 1999 average ÐÐ, N. Frank, D. Shideler, and H. M. Johnson, 1969: Atlantic trop- bias is near or below the 1990±98 average, except at ical disturbances of 1968. Mon. Wea. Rev., 97, 251±259. 72 h where it is about twice the long-term average. The Tsai, W.-Y., M. Spender, C. Wu, C. Winn, and K. Kellogg, 2000: Sea Winds of QuickSCAT: Sensor description and mission overview. 1999 absolute errors are also near or below the 1990± Proc. Geoscience and Remote Sensing Symp. 2000, Vol. 3, Hon- 98 averages, except again at 72 h where it is about 25% olulu, HI, IEEE, 1021±1023.

Unauthenticated | Downloaded 10/04/21 11:24 AM UTC